An Age Based Recovery Strategy
There is much discussion about recovery plans and the best strategy to return to normal. The CDC has proposed a phased recovery approach that we have studied in our recovery models for NYC, LA County and Georgia. Many states are moving out with this phased approach. Other states are still locked down and are just starting to slowly release. A recent model published by Penn Wharton present a binary choice between economic recovery and deaths: Penn Wharton Reopening Simulator.
There is a middle way that balances economic growth and recovery with COVID-19 mortality, an approach that allows those who are least vulnerable to get back to normal, an approach segregated by age and vulnerability. Younger age groups are allowed maximum mobility and higher age groups are restricted while the vulnerable are protected. This allows us to get the economy back on track AND minimize mortality.
We have applied our Covid Decision Model to multiple scenarios to study this approach. The results show that an age based policy for recovery, where the younger population is given maximum reasonable mobility, does work to minimize deaths. This approach allows for maximum economic activity as most manual labor and service industry jobs are populated by people under 50. People over 50 are typically able to work remotely or are retired. The assumption of continued best practices for social distancing make the contact rate in this analysis easy to achieve.
COVID-19 Infection Rates versus Age
The COVID-19 disease disproportionately affects those who have preexisting comorbidity conditions and those above 70 years of age. There is very low penetration in children and the hospitalization and death rate for those under 50 who have no preexisting comorbidity conditions is extremely low.
A recent paper published in the New England Journal of Medicine presents a detailed analysis of COVID-19 test data from Iceland. The analysis of age and susceptibility shows that the rate of infection in young people is very low. Spread of SARS-CoV-2 in the Icelandic Population
The CDC now has ample data that is consistent with the Iceland study. CDC Case Updates for the US. Children account for less than 2.5% of the positive tests, but represent approximately 25% of the overall population.
COVID-19 Death Rates versus Age
The CDC has also published extensive mortality data: Provisional COVID-19 Death Counts by Sex, Age, and State
For the period from February to early May 2020 the CDC data shows that for school children the number of deaths caused by COVID-19 is less than 0.5% of deaths. For middle age individuals COVID-19 reflects approximately 5% of overall deaths. The overall risk of death from COVID-19 is to those under 50 is very small. If we are worried about real risks, COVID-19 should be very low on the list. This article on Bloomberg illustrates this point: Coronavirus Deaths by Age: How It’s Like (and Not Like) Other Disease. If we combine the risk of infection with the risk of death, the general population mortality is exceeding low.
Protecting the Vulnerable
Those with comorbidity conditions make up close to 90% of the COVID-19 fatalities. Special care should be taken to protect these individuals. In particular, those in nursing homes or care facilities should be provided special care. The CDC has provided clear guidelines for protecting these vulnerable populations. https://www.cdc.gov/coronavirus/2019-ncov/hcp/long-term-care.html
Age Based Recovery Model
In this model different age groups are allowed to return to normal with varying degrees of social contact (connectivity). School age children are allowed to return to school. Those 50 and under are allowed to return to normal with some degree of social distancing. Those between 50 and 70 practice more social distancing and exercise a higher degree of caution. Those over 70 should continue to be cautious and minimize their social interactions. The most vulnerable continue to be protected.
Variations to Examine: Case Study Comparison
A model for variations with an R0 in the range of 3 is examined. A simulation population of 1 Million was used for all cases. Three variations of in population susceptibility and connectivity are examined, “narrow”, “nominal ” and “wide”.
Five cases for each for each distribution are examined:
- A “Wide Open” case with no mitigation (lockdown) is reference bench mark.
- A “Light Switch” model is examined to show what a full recovery to normal would look like.
- A “CDC Phased” recovery is examined in 2 month relaxation increments with all groups treated the same (except the vulnerable).
- The “Age Bin #1” case shows an aggressive age segregated recovery is examined with protection for the vulnerable.
- The “Age Bin #2” case shows a modest age segregated recovery with protection for the vulnerable.
A modest 25% reduction in mortality is assumed due to improvements in critical care.
The results of the analysis clearly indicate that the age segregated approach with variations in connectivity results in the lowest overall death rate. This is the best approach to balance economic recovery with COVID-19 mitigation.
Analysis Results: Death Rates
We examine cumulative and daily death rates for each of the cases. As indicated by the R(t), cases with higher reproduction rates generally yield higher death rates in the initial outbreak. For cases where we protect the vulnerable, the death rate will be lower for the same infection rate. For cases where we segregate by age, the death rate will be even lower for the same infection rate as those who are younger have a much lower fatality rate from COVID-19. Note that the light switch scenario does result in less overall deaths than the model for the unmitigated outbreak. This is due to the bow wave effect of an unmitigated infection. With the light switch model, there is a shift due to the distribution of the population with connectivity and susceptibility lowering the herd immunity threshold. The CDC phased approach with continued protection for the vulnerable still shows a comeback once restrictions are lifted. The age segregation approaches, Age Bin #1 and #2 cases, show a strong flattening of the death curves, even in the more aggressive Age Bin #1 case.
Daily Death Rates are shown below for reference. Note that for the Age Bin #1 and #2 cases the daily death rate is very small compared to the initial peak.
Analysis Results: Reproduction Rate over Time R(t)
The reproduction rate R(t) is examined for all three cases. R(t) over one means the infection is growing and less than one means it is decaying. If the R(t) is kept close to one, the infection grows slowly or stays relatively flat. R(t) drops as contact rates are reduced and as the susceptible population is reduced as a percentage of the whole.
The wide open reproduction rate as a function of time is shown below for all three distribution cases. Each case has been normalized to slightly over 3 for R0 (the peak at the initial outbreak). In this case the outbreak plays out and eventually dies.
R(t) unmitigated for three distributions.
The “Light Switch” model shows the suppression of the outbreak after the initial lockdown. Once things return to normal, the outbreak comes back quickly, but the peaks of R(t) is reduced as those most susceptible and most connected have been removed from the susceptible population. The simulation noise as the infection dies is due to the low number of discrete infectious agents relative to the general population.
R(t) over time for the light switch model.
We examine the R(t) for all five cases relative to the narrow, nominal and wide distribution models. In all cases the best overall effect is Age Bin #2 scenario, with R(t) held below 1. The more aggressive Age Bin #1 scenario has higher R(t), just over 1. The CDC Phased Model only holds back the infection and when activity resumes, the infection comes back, but the peak is significantly less than the light switch model.
R(t) over time for the narrow distribution.
R(t) over time for nominal distribution.
R(t) over time for the wide distribution.
Analysis Results: Vulnerable Population Infection Rates
We define the vulnerable as those with comorbidity conditions and those 70 and older. The Age Bin models are superior for protection of the vulnerable.
Analysis Results: Active Infections and Hospitalizations
Infections and hospitalization outputs match. This is indicative that hospitalization rates and uniform testing of symptomatic people will give a very clear picture of what proportion of the population is infected and what the vector of the disease is. Real data used in conjunction with a well calibrated model are effective tools for recovery management.
Cumulative Output Parameter Distributions for Wide Open (Unmitigated) Cases
Breakdown of outcome results by age for the three wide open unmitigated scenarios are shown below. Note that improbable events, like deaths in the younger age brackets, are random events in this simulation. Since we are running a simulation population of 1 million agents, low probability events may or may not occur in a given run.
Case Study Input Parameters
A simulation population of 1 million discrete agents is considered. Environmental effects are disabled. The disease timer parameters are shown below:
Three variations in connectivity and susceptibility are examined. These are labelled “Narrow”, “Nominal” and “Wide”. Please review our post on this subject to explain how this factor affects herd immunity: Are We Closer to Herd Immunity than Most Experts Say?
The time varying simulation parameters that affect contact rates (connectivity) for each case are shown below:
Wide Open Case
Light Switch Case
CDC Phased Recovery
Aggressive Age Bin #1 Recovery
Nominal Age Bin #2 Recovery
We assume 90% of deaths due to COVID-19 involve commodities. The weighted outcome table is shown below: